Introduction: Re-sorting aspectual classes

Discussions of Aktionsart and verb class generally distinguish three types of eventive VP: Incremental Theme verbs, such as eat, draw, write, and destroy, Change-of-State verbs, such as open, clear, and Xatten, and other unergative and transitive verbs, including activities, semelfactives, and some others, such as run, drool, and push. Since both Incremental Theme and Change-of-State verbs are usually accomplishments, and both may exhibit Tenny’s (1992a) measuring-out eVect with internal arguments, they have usually been treated as a natural class. This chapter shows that at least a certain subset of the third class—zero-derived denominal verbs—should also be treated as members of the Incremental Theme or Change-of-State classes.

On the l-syntactic approach of Hale and Keyser (e.g. 1993), the position of the nominal that forms the Root of the denominal verb, prior to incorpor- ation, is identical to the position of certain unincorporated measuring-out arguments. Such roots may diVer in properties that bear on measuring-out, such as inherent boundedness. Consequently, we expect that diVerent denominal verbs will have diVerent Aktionsart properties, and that such properties will be reliably determined by the meanings of their roots, in the same way that such properties aVect the Aktionsart of VP predicates with

I wish to thank the workshop organizers, Tova Rapoport and Nomi Erteschik-Shir, as well as the workshop participants and audiences at the University of Maryland and the University of Arizona, for very useful input. All remaining shortcomings are of course my own responsibility.

unincorporated measuring-out arguments. This turns out to be the case. On this analysis, however, we must assume that there are two crucially diVerent types of denominal verb in English: verbs whose names are derived via incorporation of a Root from within the argument structure, producing the measuring-out eVect, and verbs whose names are derived some other way, by a mysterious, parametrically varying, ill-understood process which I shall call ‘Manner Incorporation’.

3.2 Background

Much recent work on telicity has turned on the important connection between the direct object position and the telicity of the VP, as discussed in Verkuyl (1972), Dowty (1979; 1991), and Tenny (1992a), among many others. The central observation is that in many VPs, the boundedness of the direct object determines the telicity of the event denoted by the whole VP complex, as illustrated by the for/in temporal adverbial tests in (1) (Vendler 1957). A proposal that has gained substantial currency is that there is a functional projection which checks the boundedness features of the direct object to provide an aspectual interpretation for the VP (e.g. Borer 1998; 2002; van Hout 1996). This projection is sometimes conXated with the accusative case- checking projection, sometimes independent of it.

(1) a. Sue drank/wrote for hours/#in Wve minutes. b. Sue drank a pint of beer/wrote a story #for hours/in Wve minutes. c. Sue drank beer/wrote stories for hours/#in Wve minutes. d. Sue wrote at a story for hours/#in Wve minutes Other authors have called into question the importance of the direct object as a determiner of telicity, notably JackendoV (1991; 1996) and Levin (2000). There are verbs which take an overt, bounded, deWnite direct object and are yet inherently atelic (2a, c); they become telic when a goal argument is provided (2b, d).

(2) a. Sue pushed the cart for an hour/#in an hour. b. Sue pushed the cart to the Weld #for an hour/in an hour. c. Sue kicked the ball for an hour/#in an hour. d. Sue kicked the ball to the centre #for a second/in a second. There is a similar set of unergative verbs of motion: they are essentially atelic, as is expected since they do not have a direct object, but they may

become telic with the addition of a goal PP (still without a direct object), illustrated in (3).

(3) a. Sue danced for an hour/#in an hour.

b. Sue danced across the stage #for Wve minutes/in Wve minutes. c. Sue hopped for an hour/#in an hour.

d. Sue hopped across the stage #for Wve minute/in Wve minutes. A third class of verbs of motion may be transitive as well as intransitive, but do not become telic until a goal PP is added:

(4) a. Sue walked for an hour/#in an hour. b. Sue walked the dog for an hour/#in an hour.

c. Sue walked (the dog) to the park #for Wve minutes/in Wve minutes. With respect to verbs of motion, when motion appears to be spontaneous or internally caused, there is a well-known connection between tests for unac- cusativity (there-insertion (5), and auxiliary selection (6)) and the presence of a goal PP, implying a connection between telicity and the object position: (5) There-insertion

a. The bullet whistled as it passed my ear. b. _{There whistled a bullet (as it passed my ear).} c. There whistled a bullet past my ear.

(6) Auxiliary selection in Dutch (Borer 1998) a. Jan heeft/is gesprongen.

Jan has/is jumped ‘Jan has jumped.’

b. Jan is in de sloot gesprongen. Jan is in the ditch jumped

‘Jan has jumped into the ditch.’ where in de sloot is a Goal, not a Location

c. Jan heeft in de sloot gesprongen. Jan has in the ditch jumped

‘Jan has jumped (while) in the ditch.’ where in de sloot is a Location, not a Goal This would seem to support a necessary connection between presence of an internal argument and telicity, as predicted by measuring-out treatments, but

it is clear that it is the structural eVect of the Goal PP, rather than the telicity it can provide, that is relevant for the unaccusativity tests. Consider the Italian examples in (7):

(7) a. Gianni e` corso verso il bosco. Gianni is run towards the woods. ‘Gianni ran towards the woods.’

b. Gianni e` scivolato in direzione della pianta. Gianni is slid in the direction of the tree. ‘Gianni slid in the direction of the tree.’

Although the unaccusative auxiliary selection (e` ‘is’, as in the Dutch example in (6)) indicates that the additional PP has indeed licensed an internal argument, the PP in question in these examples does not provide an end- point, and the entire VP is atelic. Similarly, atelic PPs like towards and around license causatives of manner-of-motion verbs in English, despite the atelicity of the entire event, as shown in (8):

(8) a. John waltzed Matilda around and around the room for hours. b. John walked Mary along the river all afternoon.

Facts like these show that there is no necessary connection between the presence of the internal argument and telicity here. For a discussion of this class of verbs and its implications for treatments of Aktionsart, see Folli and Harley (2003).

A third class of atelic activity/semelfactive verbs with objects become telic only with the addition of a result phrase (Rappaport Hovav and Levin 1998): (9) a. Sue hammered the metal for Wve minutes/#in Wve minutes.

b. Sue hammered the metal Xat #for Wve minutes/in Wve minutes. c. #This metal hammers easily.

d. This metal hammers Xat easily.

Again, the presence of the internal argument is not the crucial factor in determining the Aktionsart of the VP, for these verbs.

Most theorists have ascribed the distinction between Incremental Theme verbs and the verbs discussed above to an idiosyncratic property of the verbs themselves. For example, van Hout (2000a) says of these verbs, ‘Following Dowty, Tenny, Krifka and Verkuyl, I take it that it is a lexical property of verbs that distinguishes the push-class from verbs like drink and write.’ Here, I show that these two apparently distinct classes of verb can be treated in a uniform

way, assuming an l-syntactic approach. There is an important connection between the ‘object’ position and measuring-out, as well as other argument positions which can also produce a measuring-out eVect. The crucial claim here is that in all cases the inXuence of the measuring-out argument is exerted from its base-generated position, and hence can even be seen in cases where the measuring-out argument is incorporated into the verb. A corollary of the central claim, then, is that the measuring-out argument cannot be exerting its inXuence from the speciWer of a telicity-checking functional projection. The overall view here is thus very much in the spirit of Hale and Keyser’s conclusion in this volume, that ‘[inner] aspect is orthogonal to argument structure’. It is the interaction of the meanings of their constituents and their syntactic argument structure which determines the Aktionsart properties of predicates; it is not the Aktionsart properties of predicates which determine their argument structure.